QMC Sampler...

[Dink]

The value in QMC SAMPLER

Adaptive Amount (0.1 is less quality and 1 is more quality...is right?)

Subdiv mult (0.1 also less quality , and 1 more quality)

right?

[Gijs]

Re: QMC Sampler...

adaptive amount: 0 (not recommended) means less adaptation
1 means full adaptation. In general, don't touch this setting and leave it at its default 0.85. But lower values will be the same or better quality than higher values.

subdiv multiplier: multiplies all subdiv. values in vray (except lightcache)

example: say that you have 8 subdiv in your material or light, and you have subdiv mult. of 4, then vray will take a maximum of 8x8x4 samples.

other example: say that you have irradiance hemispheric subdiv of 50, and a global subdiv. mult. of 0.5, then vray takes a maximum of 50x50x0.5 samples.

In both cases, the exact amount of samples that is taken is dependent on the noise treshold (lower numbers will make vray more sensitive to noise, thus increasing samples)

[dalomar_cg]

Re: QMC Sampler...

To add to what Gijs said,

With the adaptive amount, there's really no reason to move it much beyond .85. I believe our defaults now have it at 1, which is great for QMC calcs, DOF, and LC. But it is potentially very bad for IR. So .85 is a good base, and I would say you could increase it to about .95 for a bit more adaptation (speed). I wouldn't recommend fooling around with an amount of 1.

the subdivision multiplier also does not affect AA values or caustics subdivisions

as far as using the multiplier, I have some thoughts on the issue that I shared in another thread. Check it out if you care to
http://asgvis.com/index.php?option=c...17197#msg17197

[Gijs]

Re: QMC Sampler...

Just read that link Damien, where you mention that a multiplier of 1.5 will increase subdiv from 8 to 12 or 50 to 75. But it is not working that way.

For example: Subdiv. of 8 means 8x8 samples at most. a multiplier of 1.5 will lead to 8x8x1.5 at most, which is a subdiv of about 9.8, not 12.

[dalomar_cg]

Re: QMC Sampler...

So its working on the back end of determining the number of samples as opposed to the number of subdivisions applied. That makes sense.